Artificial Intelligence in Pharma: Revolution or “Flash in the Pan”?
Artificial Intelligence (AI) is now actively used in various industries, transforming and improving the processes efficiency. Medicine and pharmaceuticals are no exception, they are industries where AI technologies are being introduced into routine practice.
Development of new drugs is a long and expensive process; it is assumed that AI, by optimizing the processing of large amounts of data, as well as by identifying hidden patterns, etc., would significantly reduce the costs and timeframes for new drugs creation.
On April 17-18, 2024, the Big Data Age Data Fusion Conference on data analysis and artificial intelligence technologies was held in Lomonosov cluster. As part of the conference, at “ML+ Science: Pharma” session, experts discussed the application of AI and machine learning in the field of drug development. The session was moderated by Andrey Aleksandrovich Ivashchenko, Chairman of the Board of Directors of ChemRar Group.
Opening the session, Andrey Aleksandrovich noted the importance of AI and machine learning technologies in the pharmaceutical industry and asked the speakers to tell about real cases where these technologies are already successfully used, as well as where breakthroughs are expected in 5-10 years: “It is important for all of us, first of all for drug developers and investors, to understand in which areas the use of artificial intelligence technologies is efficient, and where it is just “a flash in the pan”, so as not to find ourselves in a situation like, for example, in biotechnology in the late 1990s, when the first successes in the human genome deciphering appeared. At that time, it seemed that this breakthrough would provide a complete understanding of diseases and, as a consequence, their treatment. And, huge amounts were invested in projects related to the genome deciphering. On this wave a huge venture bubble was formed, and as a result 90% of companies went bankrupt, but the remaining 10% introduced new technologies as tools in new drugs development, thanks to which we have many effective drugs and vaccines”.
Konstantin Valerievich Balakhin, Head of Innovative Drug Development at ChemRar Research Institute, LLC, Professor of MIPT, Doctor of Chemical Sciences, told about application of machine learning and artificial intelligence in drug development, in particular, in predicting the properties of low molecular weight compounds – drug candidates.
According to Konstantin Valerievich, the main strategies of drug development, which are applied in ChemRar Group, as well as globally, are de novo design, which is based on modeling a molecule from scratch; compound libraries screening – in vitro biological testing of large libraries of various compounds in specific biological tests; as well as a molecule design based on an analog, which uses the principle of previously known molecules modification. The first two technologies are innovative, high-tech and resource-intensive strategies leading to new drug chemotypes. They account for 10% of innovative drugs registered worldwide.
The third strategy (analog-based design) is the most productive rational industrial strategy; this technology produces 90% of all drugs. This strategy is being implemented by ChemRar Group on the basis of Russian universities: it is important that education and design part go simultaneously. ML/AI methods are an integral component of this technology. Konstantin Valerievich also spoke about drug development blocks, where ML/AI methods can be applied – at each stage from promising prototypes selection to patentability analysis, ML/AI technologies can significantly optimize the process. The key groups of tasks, according to Konstantin Valerievich, belong to the field of chemoinformatics and structure-property dependencies modeling. “Modern machine learning methods are a powerful practical tool in technologies for directed design of drug compounds,” summarized his presentation Mr. Balakin.
Dmitry Shkil, Chemoinformatics Manager of ChemRar Group, also spoke at the session. Dmitry spoke about the role of artificial intelligence in solving routine tasks in pharmaceuticals. ChemRar Group has an internal specially designed platform for drug discovery based on artificial intelligence, which allows solving many tasks such as: bio-target search and validation, identification of hit compounds, optimization of lead compounds, pre-clinical trials, clinical trials. At the moment, ChemRar Group is developing an LLM-based decision-making system with an ability to interact with “experts in a subfield” LLM (e.g. ADME, medicine, disease biology).